Up-regulation of HLA class-I antigen expression and antigen-specific CTL response in cervical cancer cells by the demethylating agent hydralazine and the histone deacetylase inhibitor valproic acid

C33A (HPV-), CaSki (HPV-16+) and MS751 (HPV-18+) human cervical carcinoma cell lines, which express in common the HLA-A2 allele on cell surface [26], as well as the SW480 colon carcinoma cell line, were maintained in RPMI-1640 medium (Gibco Laboratories, Grand Island, NY, USA) supplemented with 10% fetal bovine serum (HyClone, Inc. Laboratories Logan Utah. USA), penicillin G (100 IU/mL) (Lakeside, México), 100 μg/mL streptomycin sulphate (Sigma Chem. Co. St. Louis Mo, USA), and 2 mM L-glutamine (Sigma Chem. Co. St. Louis Mo, USA). Monoclonal antibodies were obtained from hybridomas supernatants and purified by elution in Protein-G sepharose (Sigma, Chem. Co. USA) columns: PA2.1 (Anti HLA-A2, -A28) MAb was purchased from American Type Culture Collection (Rockville, MD, USA) and the W6/32 MAb, which recognizes a conformational epitope on the intact heavy chain/β₂microglobulin complex, was generously supplied by Dr. Gerd Moldenhauer of the German Cancer Research Center, Heilderberg, Germany.

Cervical cancer cell lines were cultured in the presence of H, VA or both. Briefly: 5 × 10⁵ cells were cultured in 6-well plates in the presence of 10μM of H (Sigma, Chem. Co. U.S.A) or 1 mM of VA (Sigma, Chem. Co. U.S.A) during 5 or 3 days respectively or with both drugs added together. On day 3 of cell culture, 2 mL of medium were removed and then added 2 mL of fresh complete medium containing the same concentration of drugs.

To determine HLA class-I molecule expression on cell surface, 5 × 10⁵ cells were treated with 10 μg/mL of each purified MAb, for 30 min. After washing the cells three times in 0.15 M NaCl-0.01 M phosphate buffer (pH 7.2)-2% fetal bovine serum (PBS-F), FITC-labeled goat anti-mouse Ig antibody (Gibco Co. U.S.A) was added to a dilution of 1:100 for 20 min on ice, followed by another two washes in PBS-F. Finally, the cells were resuspended in 0.5 mL of PBS-F and 1 μg/mL of Propidium Iodide (Sigma Chem. Co. USA) to discard cellular debris and then transferred to tubes. Cell samples were analyzed in a FACS-calibur flow cytometer (Becton Dickinson & Co. Mountain View, CA, USA). After gating out cell debris, 10 000 events were analyzed for their fluorescence intensity. In all experiments, the fluorescence intensity was determined at least three times where each of the 10, 000 events were gated. The staining with the FITC-labeled secondary antibody alone was considered as a negative control. The effect of human recombinant IFN-gamma on the HLA induction was determined in cell lines cultured with or without the presence of 200 U/mL of IFN-gamma (Genzyme Diagnostics, Cambridge, USA) for 48 hours [27]. The cells were then harvested and their HLA expression was determined as previously indicated.

Elution buffer. The PCR-amplification was carried out using the Advantage-GC Genomic PCR Kit (Clonetech) according to the manufacturer's instructions. PCR primer sequences for amplifying the human MHC class I promoter are enlisted in table 1. PCR products were separated on a 1.8% agarose gel and visualized by ethidium bromide staining.

The antigenic peptides TLGIVCPIC and YMLDLQPETT derived from the E7 HPV-16 protein and the KLPDLCTEL derived from the E6 HPV-18 protein that specifically bind to HLA-A2 allele [28‐30], were synthesized by Invitrogen, USA, dissolved in phosphate buffered saline and stored at -70°C before use. The peptide GILGFVFTL derived from the M Influenza-A protein, was used as positive control for binding assay to the empty HLA-A2 molecule. Lymphoblastic T2 cell line, which express empty HLA-A*0201 molecules on its cell surface [31], was used to test the affinity of different concentrations (12.5–100 μM) of each synthetic peptide after incubation overnight at 37°C in the presence of 5 μg/mL of β₂-microglobulin (Sigma, USA).

Biopsies were taken from areas with visible macroscopic cervical tumor using a sterile biopsy punch. Part of the biopsy was sent to the Institution's Pathology Department for routine hematoxilin & eosin diagnosis. The remaining biopsy specimen was immediately frozen at -20°C for HPV typing. In addition, 20 mL of peripheral blood were drawn from the arm by venipuncture to obtain the mononuclear cell fraction in order to stimulate the cytotoxic T lymphocytes. The protocol was approved by the Institutional Regulatory Boards and patients signed an informed consent before blood and sample tissues were taken.

The MY09 and MY11 L1 consensus primers (Table 1) that recognize a conserved region in the L1 open reading frame, producing a fragment of 450 bp, were used to examine the presence of HPV DNA in the genomic DNA of each β-globin positive tumor sample.

The reaction was carried out in a final volume of 25 μL containing 400 ng of DNA, 1.5 mM MgCl₂, 200 μM of dNTPs, 0.4 μM of each of the primers and 1U of Taq DNA polymerase (GIBCO BRL, Grand Island, NY). The positive control consisted of DNA from CaSki and MS751 cell lines, which contain the HPV type 16 and 18 genome respectively. The conditions of amplification were as follows: Denaturing at 94°C for 15 sec, primer annealing at 58°C for 30 sec and extension at 72°C for 1 min, for a total of 35 cycles, the final cycle included an incubation at 72°C for 10 min. 7 μL of amplification product were electrophoresed in 1.5% agarose containing 0.5 μg/mL of ethidium bromide and visualized by UV light. Positive MY09/MY11 products were digested with Bam HI and Rsal restriction enzymes (Biolabs, USA). The restricted samples were electrophoresed on a 3% agarose gel stained with ethidium bromide. The restriction fragment length polymorphism (RFLP's) obtained were compared with that reported by Bernard [32].

To stimulate CTLs, we used a method previously reported [33]. Briefly: 4 × 10⁶ Peripheral Blood Lymphocytes (PBLs) were resuspend in 1 mL of complete medium consisting of Iscove's Modified Dulbecco's Medium (Gibco BRL, USA) supplemented with 10% heat-inactivated FBS, 100 IU/mL penicillin, 4 mM L-glutamine, 1 mM sodium pyruvate and 20 μM 2-mercaptoethanol, and incubated with 10 μM of peptide in 24-wells plates. On day 3, the wells were topped up with 1 mL of complete medium containing recombinant human IL-2 (final concentration 10 IU/mL, R&D). On day 7 and weekly thereafter, the cells were restimulated as follows: (a) we used T2 cell line as antigen presenting cell; 1 × 10⁵ T2 cells previously loaded with 50 μM of the peptides in the presence of β₂-microglobulin and fixed with 0.1% glutaraldeyde in PBS, were incubated with 5 × 10⁵ T cells; (b) 1 × 10⁶ responder T cells were added in 1 mL of complete medium; and (c) cells were topped up 2 days later with 1 mL of complete medium containing hrIL-2 and hrIL-15 at final concentration of 10 IU/mL and 15 ng/mL respectively. Cytotoxicity assays were performed on day 21.

Cervical cancer cell lines alone or pretreated with H, VA, both, IFN-gamma or H/VA/IFN-gamma as indicated, were used as target cells after labeled with ⁵¹Cr (Amersham, USA) for 1 h. Different numbers of effector cells in 50 μL of complete medium were incubated and then 2.5 × 10^{3 51}Cr-labeled target cells were added to triplicate wells of 96-well plates in final volume of 200 μL. After 4 h at 37°C, 100 μL of supernatant were harvested and transferred to counting vials and measured on a γ-counter (Packard, U.S.A). For each pretreated cell group, ⁵¹Cr-labeled cells incubated with 5% SDS or medium alone were used to determine maximum and spontaneous releases. Spontaneous release was usually less than 10% and never exceeded 15%. The percentage of specific lysis of each well was calculated as: (experimental release - spontaneous release)/(maximal release - spontaneous release) × 100.

All numerical data were expressed as average of values obtained ± standard deviation (SD) of experiments made by triplicate. Comparisons were evaluated by unpaired t test. A p value <0.05 was considered significant.

To determine whether these epigenetic agents enhance the constitutive expression of HLA class-I molecules, the expression analysis of the HLA-A2 allele and total HLA class-I molecules was carried out by using PA2.1 and W6/32 MAbs. The results showed that HLA-A2 allele expression level was unchanged in the C33A cells by hydralazine alone whereas VA, H/VA, IFN-γ and H/VA/IFN-γ increased one-fold its expression. Regarding total class-I molecules, the increasing effect was unexistent except for a small increase by IFN-γ and H/VA/IFN-γ. In CasKi cells, a similar pattern of increased expression was observed in HLA-A2 allele and total HLA class-I molecules expression by these drugs and combinations except for hydralazine alone treatment. In particular for total HLA class-I, it seems there was a summatory effect among the three drugs, H/VA/IFN-γ. Of note the effect seen on CasKi cells in HLA-A2 allele and total HLA class-I molecules by these drugs and combinations was almost identical in the MS751 cells. Statistical significance among cell lines and treatments in comparison to untreated (wt) are shown (Figure 1).

To investigate whether the up-regulating effects of these drugs of HLA class-I molecules as shown by flow cytometry could be mediated by increased transcription, treated cell lines were analyzed by RT-PCR. Figure 2 shows that C33A cells despite had no increase in transcript levels for the HLA-A and -C genes with any combination of treatments, HLA-B gene showed a 0.35, 0.29, 0.21 and 0.42 fold-increase in band intensities with H, VA, H/VA and H/VA/IFN-gamma respectively. In CasKi cells where HLA-A2 was most increased by IFN-gamma and H/VA/IFN-gamma the fold increases in band intensity were 0.13 and 0.91 respectively. HLA-B was also increased 0.12, 0.43 and 0.28-fold with H/VA, IFN-gamma and H/VA/IFN-gamma respectively. In HLA-C, an increase of 0.25 and 1.4-fold were observed with IFN-gamma and H/VA/IFN-gamma. The MS751 cell line showed increases of the same magnitude in band intensities with all the combinations except for H alone. In particular for HLA-A gene, the triple combination of H/VA/IFN-gamma led to a 1.29-fold increase.

Previous studies have demonstrated that epigenetic mechanisms are main regulators of the expression of this class of molecules and that both DNA methylation and HDAC inhibitors demethylate and reactivate their expression. To investigate this issue, we determined by methylation-specific PCR the methylation status at the gene promoter of HLA-A, -B and -C genes in C33A, CasKi and MS751 cell lines. As shown in figure 3a, there was complete demethylation at these three promoters in all the cell lines investigated. The absence of gene promoter at these genes prompted us to analyze whether histone acetylation could be responsible for the increase expression seen by the epigenetic drugs used. As shown in figure 3b, chromatin immunoprecipitation assay showed that the combination of H/VA but no IFN-γ led to H4 hyperacetylation at the HLA-class-I promoter. Because hydralazine can be considered as a weak DNA methylation inhibitor and it has been reported that 5-aza-2'-deoxycytidine does demethylate the HLA-B promoter in the KYSE esophageal carcinoma cell line, we searched the expression of HLA-A, -B and -C genes and the promoter methylation status in several cell lines. We found that the SW480 colon carcinoma cell line had methylated the HLA-B locus. When this cell line was treated with H, VA and H/VA, like to that observed for cervical cancer cell lines, VA and H/VA led to small but clear increase in expression level of the three loci, however, neither H nor 5-aza'-2'-deoxycytidine demethylated the HLA-B locus (Figure 3c and 3d).

To analyze whether the treatment of cervical cancer cells with hydralazine and valproic acid is also able to increase their immune recognition, T lymphocytes derived from cervical cancer patients with HPV-16 or HPV-18 infection and with the HLA-A2 allele in their HLA Class-I haplotype, were stimulated with three known E6 and E7 HPV derived antigenic peptides, that specifically bind to the HLA-A*0201 allele [28‐30]. Two of the peptides TLGIVCPIC and YMLDLQPETT were derived from the E7 HPV-16 protein and the other one KLPDLCTEL derived from the E6 HPV-18 protein. We also used the lymphoblastic T2 cell line to stimulate T lymphocytes contained in PBL's from patients with cervical cancer. Due to the fact that the T2 cells express empty HLA-A2 molecules on their cell surface [31], we previously performed peptide binding assays to analyze the binding affinities for these peptides. Using 50–100 μM of these three peptides, we observed an efficient stabilization of the HLA-A2 allele on T2 cells similar to the one obtained with the control peptide GILGFVFTL derived from the protein M of the influenza-A and with high binding affinity to the HLA-A2 allele (Figure 4). The T lymphocytes used were obtained from four patients with cervical squamous cell carcinoma. Two of those with HPV-16 infection (Patients 1, 2) and two with HPV-18 infection (Patients 3 and 4) all positive for the HLA-A*0201 allele (Table 2). The lymphocytes were stimulated during three rounds with the T2 cells loaded with the three antigenic peptides and then challenged against CaSki or MS751 cells that were previously treated with H, VA, H/VA, IFN-gamma and H/VA/IFN-gamma. We observed as expected, that T lymphocytes from the patients 1 and 2, that were positive for HPV-16 infection and stimulated with T2 cells loaded with the peptides TLGIVCPIC and YMLDLQPETT were able to lyse CaSki cells (Figure 5) and that this cytotoxicity mainly increased when the cells were previously treated with VA, H/VA, IFN-gamma and H/VA/IFN-gamma. Of note cytotoxicity was at least if not higher with any of these combinations as compared to IFN-gamma alone. On the other hand the T lymphocytes derived from the two patients with HPV-18 infection (Patients 3 and 4) and stimulated with the T2 cell line loaded with the peptide KLPDLCTEL, were able to lyse MS751 cells. In patient 3, the higher cytotoxicity was found with VA, H/VA and H/VA/IFN-gamma whereas in patient 4, the cytotoxic effect on cells treated with H/VA, IFN-γ and H/VA/IFN-gamma was essentially of the same magnitud but higher than IFN-gamma alone. (Figure 5). In all experiments T lymphocytes stimulated with the E6 and E7 epitopes were always capable to lyse the T2 cell line loaded with the proper antigenic peptide (data not shown). Moreover, we observed a very low T cell cytotoxic activity on CaSki and MS751 cells when T lymphocytes previously stimulated with the control peptide GILGFVFTL were challenged against these cells

To investigate whether these epigenetic agents modulate the expression of E6 and E7 genes in the Caski and MS751 cell lines, the expression of these genes was analyzed by RT-PCR. The results show that neither E7 transcript of the HPV-16 nor E6 transcript of the HPV-18 were changed by drug treatment suggesting that the enhanced immune recognition of CaSki and MS751 cells by CTLs derived from cervical cancer patients can be mainly due to the increased presentation of antigenic peptides by the increased expression of HLA class-I molecules on cell surface rather than by an increase in E6 or E7 peptides (figure 6).